ARTHRITIS & RHEUMATISMVol. 60, No. 5, May 2009, pp 1261–1271DOI 10.1002/art.24498© 2009, American College of Rheumatology

Transmembrane BAFF From Rheumatoid SynoviocytesRequires Interleukin-6 to Induce the Expression ofRecombination-Activating Gene in B Lymphocytes

Caroline Rochas,1 Sophie Hillion,1 Alain Saraux,1 Rizgar A. Mageed,2 Pierre Youinou,1

Christophe Jamin,1 and Valerie Devauchelle1

Objective. B cells that accumulate in the synovialtissue of rheumatoid arthritis (RA) patients revise theirreceptors due to coordinate expression ofrecombination-activating gene 1 (RAG-1) and RAG-2genes. The aim of this study was to determine themechanisms that control this re-expression.

Methods. B cells from healthy control subjectswere cocultured with fibroblast-like synoviocytes (FLS)from patients with RA and osteoarthritis (OA). Re-expression of RAG messenger RNA (mRNA) and pro-teins was analyzed by reverse transcription–polymerasechain reaction (RT-PCR) and indirect immunofluores-cence. Activity of RAG enzymes was evaluated by flowcytometry to measure variations in immunoglobulin �and � light chain expression and by ligation-mediated–PCR to assess specific DNA breaks. Blocking antibod-ies, short hairpin RNA, and recombinant cytokine wereused to identify the molecules involved in RAG re-expression.

Results. RA FLS, but not OA FLS, induced B cellsto re-express RAG mRNA and proteins. Enzymes werefunctional, since the �-to-� ratios decreased and spe-cific DNA breaks were detectable after coculture withRA FLS. Transmembrane BAFF provided the first sig-

nal of RAG re-expression, since its down-regulation inRA FLS prevented RAG gene transcription in B cells.The failure of transmembrane BAFF from OA FLS toinduce RAG suggests that a second signal was providedby RA FLS. Interleukin-6 (IL-6) is a candidate, sinceblockade of its receptors precluded transcription ofRAG genes by RA FLS. Unless supplemented with IL-6,OA FLS were unable to induce RAG gene expression innormal B cells.

Conclusion. Two independent signals are re-quired for the induction of RAG gene expression in Bcells that infiltrate the synovium of patients with RA.

Rheumatoid arthritis (RA) is a common inflam-matory autoimmune disease (1) initiated by inflamma-tion of the synovial tissue (ST). A cascade of ensuingevents then leads to the formation of pannus. Thisstructure contains proliferating fibroblast-like synovio-cytes (FLS) and dividing T and B lymphocytes (2). Thelatter cells produce antibodies with specificity for the Fcregion of IgG (3) or antibodies to cyclic citrullinatedpeptides (4). They also release proinflammatory cyto-kines, such as tumor necrosis factor � (TNF�) (5) andinterleukin-6 (IL-6) (6).

Activation of ST lymphocytes is likely to occurthrough locally triggered mechanisms (7). In this regard,BAFF may substitute for the role of T cells in activatingB cells (8). To rescue B cells from apoptosis, BAFF actsin a transmembrane form or as a cell-free ligand (9). Itsreceptors include TACI, the B cell maturation antigen,and the so-called BAFF receptor 3 (B lymphocytestimulator receptor 3 [BR-3]) (10). Once recruited, Bcells spread throughout the ST or form clusters in theform of germinal center–like aggregates (11).

Instead of depending on this cytokine-mediatedsignaling, an alternative way in which autoreactive B

Supported by the Association de Recherche sur la Polyarth-rite.

1Caroline Rochas, BSc, Sophie Hillion, PhD, Alain Saraux,MD, PhD, Pierre Youinou, MD, DSc, Christophe Jamin, PhD, ValerieDevauchelle, MD, PhD: Universite Europeenne de Bretagne, Univer-site de Brest, IFR 148 ScInBioS, and Centre Hospitalier Universitaire,Brest Hopital Morvan and Cavale Blanche, Brest, France; 2Rizgar A.Mageed, PhD: William Harvey Research Institute, Barts and theLondon Queen Mary School of Medicine and Dentistry, London, UK.

Address correspondence and reprint requests to Pierre Youi-nou, MD, DSc, Laboratory of Immunology, Brest University MedicalSchool Hospital, BP 824, F 29 609 Brest, France. E-mail: youinou@univ-brest.fr.

Submitted for publication August 22, 2008; accepted inrevised form February 12, 2009.

1261

lymphocytes escape apoptosis is by altering their anti-body specificity. In particular, the B cell receptor (BCR)of immature B lymphocytes could be edited (12) todispose of high-affinity autoantibody specificities (13).There is additional evidence that mature B cells mayundergo V(D)J gene editing in the germinal centers of aperipheral lymphoid organ (14) in order to offset thedecrease in BCR avidity that occurs because of thestochastic nature of hypermutation. All of these eventsare presumed to work ultimately toward rescuing B cellsthat develop nonfunctional BCRs.

In contrast to editing, revision is turned off byengagement of the BCR with its target antigen (15). Forthis to occur, recombination-activating gene 1 (RAG-1)–and RAG-2–encoded enzymes must be expressed in acoordinated manner (16). B cells that accumulate in theRA joint re-express RAG (17) and revise their receptorsinside (18) the germinal center, but B cells can alsorevise their BCRs outside the germinal center (19).Again, given that BCR revision is driven by the need toincrease diversity (15,16,20,21), self reactivity may begenerated inadvertently, and a breach of self tolerancemay result from the ensuing defective regulation.

The aim of the present study was to gain insightinto the mechanisms that control the re-expression ofRAG in B cells derived from ST obtained from patientswith RA. We hypothesized that replacement of antibodyV genes could be a mechanism through which BAFFpromotes B cell survival. We found evidence of theinvolvement of FLS-derived BAFF in RAG gene re-expression. However, engagement of transmembraneBAFF on FLS in cognate interaction with B cells was notof itself sufficient to trigger RAG gene expression. FLSfrom patients with osteoarthritis (OA) also expressedtransmembrane BAFF, yet they were unable to influ-ence the outcome of RAG gene expression on B cells.Once prompted by FLS-mediated signals, synovial Bcells required a second signal, which was generated byIL-6. Similar combinations of factors have been reportedto induce RAG expression in immature (22) and mature(23) B cells, most notably in systemic lupus erythemato-sus (SLE) B cells (24). The current data indicate that Bcells require 2 separate signals to induce RAG geneexpression, the first from transmembrane BAFF on RAFLS and the second from one or more cytokines,including IL-6.

MATERIALS AND METHODS

Patients. ST samples were collected from 9 patientswith RA and 7 patients with OA who were undergoing knee

joint replacement surgery. Paired samples of blood were alsoobtained from 5 of the 9 RA patients. The OA patients servedas disease controls for the ST samples, and 4 laboratory staffserved as normal controls for the blood samples. The RApatients had not received any medication injected into thesynovium for at least 1 month prior to sampling. All RApatients fulfilled the American College of Rheumatologycriteria for the disease (25). OA was diagnosed according toclinical features. The 697 pre–B cell line cells generouslydonated by Dr. Paul Guglielmi (INSERM, Montpellier,France) were used as positive controls for RAG expression.

Culture of FLS. ST samples were carefully sheared outto avoid contamination with other tissues. The ST pieces werethen processed to a single-cell suspension by digesting thetissue for 2 hours at 37°C with 4 mg/ml of collagenase/Dispase(Sigma-Aldrich, St. Quentin Fallavier, France) in phosphatebuffered saline containing Dulbecco’s modified Eagle’s me-dium (DMEM). The cells were collected in DMEM (Gibco,Cergy-Pontoise, France) supplemented with 10% fetal calfserum and antibiotics. After a 48-hour culture, the nonadher-ent cells were discarded, and the adherent cells weretrypsinized. Cells were passaged only twice to reduce pheno-type and functional changes caused by repeated passaging(26). All cells were positive for CD90 and negative for HLA–DR, as determined by immunofluorescence and fluorescence-activated cell sorting (FACS).

Isolation of B cells from blood. Blood mononuclearcells were isolated by Ficoll-Hypaque centrifugation. T cellswere rosetted, and enriched B cells were recovered. Cells werestained with fluorescein isothiocyanate (FITC)–conjugatedanti-CD19 plus phycoerythrin (PE)–conjugated anti-CD5monoclonal antibodies (mAb), and the CD5–CD19� B cellswere analyzed by FACS. The cell preparations were shown tobe �98% CD19�.

Coculture of B cells with FLS. FLS were seeded in24-well flat-bottomed culture plates and grown to confluence.Then, 1–5 � 105 B cells in 1 ml of DMEM were added. Toavoid allogeneic responses and long-term culture artifacts,supernatants and B cells were collected separately only 2 dayslater. Viability of the B cells was assessed by Trypan blueexclusion. In some experiments, 10 �g/ml of anti-IgM–coatedbeads, 1 �g/ml of recombinant BAFF (PeproTech, Le Perray-en-Yvelines, France), 40 ng/ml of anti–IL-6 receptor (anti–IL-6R) mAb (R&D Systems, Lille, France), 4 ng/ml of IL-6(ImmunoTools, Friesoythe, Germany), IL-4 (ImmunoTools),or granulocyte–macrophage colony-stimulating factor (GM-CSF; R&D Systems) were added prior to the beginning ofcoculture. Each optimal concentration was determined inpreliminary experiments.

Transwells (Corning, Corning, NY) were used to eval-uate the requirement for cell–cell contact. Confluent FLS wereplaced in the lower chamber and B cells in the upper. In someexperiments, RA FLS were treated with 50 �g/ml of mitomycinC (Sigma-Aldrich) for 30 minutes at 37°C prior to theircoculture with B cells.

FACS analyses. FITC- and PE-conjugated anti-CD19,PE-conjugated anti-CD5, PE-conjugated anti-CD40, PE-conjugated anti-CD90, and FITC-conjugated anti–HLA–DRmAb, as well as FITC-conjugated annexin V were fromBeckman Coulter (Villepinte, France). PE-conjugated anti-CD40L was from BD Biosciences (Le Pont de Claix, France),

1262 ROCHAS ET AL

and PE-conjugated anti-� and FITC-conjugated anti-� werefrom Dako (Trappes, France). Anti-BAFF (Research Diag-nostics, Flanders, NJ) and anti–BR-3 (CliniSciences, Mon-trouge, France) were revealed by treatment with FITC-conjugated anti-mouse IgG antibodies (JacksonImmunoResearch, Newmarket, UK). All analyses were per-formed using an Epics XL FACS instrument (BeckmanCoulter).

Messenger RNA (mRNA) isolation and reversetranscription–polymerase chain reaction (RT-PCR) analysis.The RNAble method (Eurobio, Paris, France) was used toextract mRNA, which was then reverse-transcribed in 20 �l ofSuperScript II RNase H reverse transcriptase (Invitrogen,Cergy-Pontoise, France). RAG-1 and RAG-2 mRNA wereamplified by nested RT-PCR using 1 �l of complementaryDNA (cDNA), with the primer pairs described elsewhere (22)and Taq DNA polymerase (Invitrogen). In the first round ofPCR, cDNA was amplified for 25 cycles of 30 seconds at 94°C,1 minute at 56°C, and 1 minute at 72°C, with a final 10-minuteextension at 72°C. The second round of PCR was for 35 cyclesusing the same conditions. There was 1 round of a 40-cyclePCR for GAPDH RT-PCR amplification. The final productswere visualized on SYBR Green–stained 1.5% agarose gels.

Isolation and analysis of single cells. B lymphocyteswere deposited into PCR tubes containing 10 �l of RT buffersupplemented with 1� first-strand buffer (Invitrogen), 5 �Mrandom hexamer primers, 0.01M dithiothreitol, and 0.5 mMdNTPs (all from Promega, Charbonnieres, France). IndividualB cells were sorted using an Epics Elite FACS instrumentoutfitted with an Autoclone single-cell deposition unit (Beck-man Coulter). RT and nested PCR for mRNA for the RAG-1,RAG-2, and GAPDH genes were performed as describedpreviously (22–24). Briefly, after mRNA conversion for 2 hoursat 42°C with SuperScript II RNase H reverse transcriptase, afirst round of PCR for 25 cycles was performed as describedabove, which was followed by a second round for 40 cycles.

Ligation-mediated (LM)–PCR. DNA was extractedfrom 105 cells using a DNAzol kit (Invitrogen), and 500 ng wasligated to the LM-PCR linker at 20 pmoles using T4 DNAligase (Promega). After a denaturation step at 94°C for 15minutes, a first round of PCR for 15 cycles of 40 seconds at94°C, 1 minute at 72°C, and a 10-minute extension at 72°C wasfollowed by a 3-minute denaturation at 94°C and a secondround of PCR for 40 cycles using previously described primers(22) lying upstream of the 5� end of either the J�3–J�7 or theJ�2–J�5 genes were paired with a primer encompassing theLM-PCR linker and the recombination signal sequence. ThePCR products were then analyzed on 2% agarose gels stainedwith ethidium bromide. Their specificity was confirmed bysequencing.

Measurement of cytokines in FLS supernatants. Theharvested supernatants were first centrifuged, and IL-6 wasimmediately assayed using a commercial enzyme-linked immu-nosorbent assay (ELISA) kit according to the manufacturer’sinstructions (Beckman Coulter). The level of BAFF wasdetermined using an in-house ELISA (27).

Immunofluorescence staining of B lymphocytes andFLS. B cells were collected from the cultures and prepared asdescribed elsewhere (22). Cells were stained first with rabbitanti–RAG-1 antibody (Santa Cruz Biotechnology, Santa Cruz,CA) followed by FITC-conjugated anti-rabbit IgG (Jackson

ImmunoResearch) and stained second with goat anti–RAG-2antibody (Santa Cruz Biotechnology) followed by biotinylatedanti-goat IgG (Jackson ImmunoResearch) plus tetramethyl-rhodamine isothiocyanate (TRITC)–streptavidin. As negativecontrols, cells were similarly stained, but without the first layerof anti-RAG antibody. The cells were analyzed with a LeicaTCS-NT confocal microscope (Leica, Wetzlar, Germany).Percentages of double-positive cells were determined in 3randomly selected fields by 3 independent observers (CR, SH,and CJ).

After 2 passages, FLS were mounted onto poly-L-lysine–coated slides and incubated with anti-BAFF or anti–BR-3 mAb followed by FITC-conjugated anti-mouse antibody(Jackson ImmunoResearch). After washes, the cells were fixedand analyzed as above. Control mouse IgG did not produceany fluorescence background when developed with FITC-conjugated anti-mouse antibody.

Short hairpin RNA (shRNA) interference. RA FLSwere transiently transfected with the shRNA sequence 5�-TGTGGAAAGGACGAAACACC-3� (RHS 4430 and RHS4477; Open Biosystems, Huntsville, AL) using the pGIPZhuman lentiviral vector. The cells were grown in 24-well dishesand transfected at a 5:1 ratio in RPMI 1640 according to themanufacturer’s instructions. Aliquots of cells cultured withscrambled plasmids (5�-ATCTCGCTTGGGCGAGAGTAA-G-3�) were used as controls. Knockdown of BAFF was as-sessed by FACS, and the results were expressed as percentagesof positive cells. The mean fluorescence intensity (MFI) ofBAFF in transfected cells relative to nontransfected cells wasalso measured.

Figure 1. Expression of recombination-activating gene 1 (RAG-1)and RAG-2 in B lymphocytes. A, B cells from paired blood andsynovial tissue (ST) samples from 5 patients with rheumatoid arthritis(RA) were sorted, and transcripts of RAG-1, RAG-2, and GAPDHgenes were amplified. The 697 pre–B cell line served as a positivecontrol for RAG gene expression throughout the study. B, Single ST Bcells from 3 of the 5 RA patients (RA1, RA2, and RA5) were sortedusing the Autoclone single-cell deposition unit, and transcripts ofRAG genes were individually obtained. Results for B cells from patientRA1 are shown. C, Frequency of RAG-1� and/or RAG-2� individualB lymphocytes was determined. Values are the mean and SD of 3experiments.

BAFF AND IL-6 FROM RA SYNOVIOCYTES INDUCE RAG IN B CELLS 1263

RESULTS

In vivo RAG expression. B cells in the ST ofpatients with RA and in the blood of patients with SLEaberrantly express RAG-1 and RAG-2 enzymes (17,19).To identify mechanisms that promote RAG gene tran-scription in ST B cells, we determined the presence ofRAG-1 and RAG-2 transcripts in B cells from the bloodand ST of 5 RA patients. Whereas mRNA for RAG-1was detected in blood B cells from only 1 of the patients(RA2), mRNA for both RAG-1 and RAG-2 genes weredetectable in the paired samples of ST B cells from 4 ofthe 5 patients (RA1, RA2, RA3, and RA5) (Figure 1A).

Single-cell RT-PCR for RAG-1 and RAG-2mRNA made it possible to demonstrate their synchro-nized transcription in individual B lymphocytes. Based

on CD19 expression, B lymphocytes were single-cellsorted from the 5 RA ST samples, and their RAGmRNA amplified by RT-PCR. We restricted this analy-sis to B cells containing GAPDH mRNA, but notgenomic DNA. Transcripts for RAG-1 and RAG-2genes (lanes 5, 9, and 19 in Figure 1B) coexisted in amean � SD of 13.3 � 1.4% of B lymphocytes from theST samples (Figure 1C). These data suggest that thesignals that trigger the V(D)J recombination machineryare provided by the ST environment.

In vitro regulation of RAG expression. Inductionof RAG mRNA by coculture of normal blood B cells withRA FLS. FLS from 9 patients with RA, including the 5studied in the previous experiment, and FLS from 7patients with OA were cocultured with blood B cells

Figure 2. Expression of recombination-activating gene 1 (RAG-1) and RAG-2 in blood B cellscocultured with fibroblast-like synoviocytes (FLS). A, RAG-1 and RAG-2 gene mRNA in B cellsfrom the blood of healthy controls were amplified by nested reverse transcription–polymerase chainreaction before and after a 2-day coculture with FLS from 9 rheumatoid arthritis (RA) or 7osteoarthritis (OA) patients, with or without 10 �g/ml of anti-IgM–coated beads. Results using Bcells from healthy control donors 1 and 2, which were tested in the presence of anti-IgM (of the 4healthy controls tested without anti-IgM), are shown. The 697 pre–B cell line served as a positivecontrol for RAG gene expression throughout the study. B, RAG-1 and RAG-2 mRNA in B cellsfrom the blood of patient RA3 were amplified before and after a 2-day coculture with a pairedsample of FLS. C, B cells from 3 healthy controls were cocultured with FLS from 3 RA patients.After 2 days, individual B cells were sorted, and mRNA for RAG-1 and RAG-2 was analyzed andthe frequencies of RAG-1� and RAG-2� transcripts were determined. Values are the mean andSD. D, B cells were cocultured with RA FLS. After 2 days, RAG-1 and RAG-2 proteins werestained with fluorescein isothiocyanate– and tetramethylrhodamine isothiocyanate–conjugatedantibodies, respectively, and the frequencies of RAG-1� and RAG-2� cells were determined in 3randomly selected fields in each of 3 separate experiments. Values are the mean and SD.

1264 ROCHAS ET AL

from healthy controls. RA FLS up-regulated mRNA forboth RAG-1 and RAG-2 genes in the B cells (Figure2A). In contrast, OA FLS did not lead to RAG up-regulation. B cell samples from 4 different healthydonors yielded similar results. Both RAG genes wereup-regulated in the presence of FLS from each of the 9different RA patients, but not with any of the 7 differentOA patients. Additionally, blood B cells from patientRA3 cultured with FLS from the same patient resultedin mRNA up-regulation of both RAG genes (Figure2B), indicating that the response with B cells from thehealthy donors was not due to an allogeneic reaction.

Turn off of RAG expression upon BCR engage-ment. Based on previous studies showing that strongBCR engagement interrupts the recombination ofV(D)J genes by turning off RAG gene expression (15),we attempted to verify that stimulation with anti-IgM atthe beginning of culture reverses RAG gene expression.As expected, RAG gene transcription triggered by RA

FLS alone was negated by concurrent BCR engagement(Figure 2A). Although RAG-2 gene expression alonepushes activated B cells into apoptosis (28), these B cellswere still alive in the coculture with RA FLS, since theydid not stain with annexin V (data not shown).

Expression of RAG-1 and RAG-2 genes at single Bcell levels. New V(D)J rearrangement indicates thatRAG-1 and RAG-2 are coordinately expressed (16). Toensure that transcripts of both RAG genes were presentin a given cell at the same time, individual B lymphocytesfrom 3 normal donors were sorted after coculture withFLS from a different RA patient for each and examinedfor the presence of transcripts for the RAG-1 andRAG-2 genes. Again, only B cells containing mRNA forGAPDH, but not genomic DNA, were analyzed. RT-PCR revealed that a mean � SD of 17.5 � 4.3% of Blymphocytes contained mRNA for RAG-1 and RAG-2genes (Figure 2C). No RAG induction was seen in OAFLS cocultures (P � 0.005 compared with RA FLS).

Figure 3. Recombination-activating gene (RAG) activity in B cells from the blood of healthycontrols cocultured with fibroblast-like synoviocytes (FLS) from 4 rheumatoid arthritis (RA) or 3osteoarthritis (OA) patients. A, Ratios of � to � were calculated in B cells before and after coculturewith RA or OA FLS. B, Percentages of fluorescein isothiocyanate (FITC)–conjugated anti-�� andphycoerythrin (PE)–conjugated anti-�� B cells before and after a 2-day coculture were deter-mined. Representative examples of 4 cocultures with RA FLS and 3 cocultures with OA FLS areshown. Broken line indicates isotype control. C, B cells were labeled with 10 nM carboxyfluoresceinsuccinimidyl ester (CFSE), cocultured for 2 days with RA FLS or OA FLS, and their proliferativeresponses were evaluated by flow cytometry. Broken line indicates CFSE staining before coculture.D, RAG-1 and RAG-2 activities in B cells after 2 days of coculture with RA or OA FLS wereevaluated by ligation-mediated–polymerase chain reaction with different sets of specific primers todetect J� and J� signal breaks.

BAFF AND IL-6 FROM RA SYNOVIOCYTES INDUCE RAG IN B CELLS 1265

In addition, fluorescence analyses establishedthat both RAG proteins were detectable. Their coexist-ence was evidenced by the yellow color when the FITC-labeled RAG-1 image was overlaid on the TRITC-labeled RAG-2 image (results not shown). Thesefluorescence stainings were undetectable in normal Bcells cocultured with OA FLS. After incubation with RAFLS, 19.6 � 0.7% of normal B cells expressed RAG-1and RAG-2 proteins (Figure 2D).

Functionality of RAG-1 and RAG-2 proteins. Us-ing the frequency of � light chain expression as anindicator of secondary gene rearrangement, we definedan index of recombination activity in the B cells (21).Accordingly, the increase in the usage of �, relative tothat of �, and hence, changes in the ratio of � to �, weremeasured before and after coculture (Figure 3A). A

2-day exposure of B cells to FLS from 4 RA patientsreduced �-to-� ratios (P � 0.05) from a mean � SD of1.3 � 0.1 to 0.6 � 0.1 (Figures 3A and B). In contrast,these ratios remained constant at 1.2 � 0.1 when the Bcells were cocultured with FLS from 3 OA patients(Figures 3A and B). Changes in �-to-� ratios due topreferential proliferative responses of a subset of B cellswere excluded, since no proliferation could be detected(Figure 3C).

To strengthen the data concerning revision of theBCR in the presence of RA FLS, we assessed thespecific RAG-dependent DNA breaks in the recombi-nation signal sequence using LM-PCR. In the absence ofRA FLS, there were no LM-PCR products in the B cells(data not shown). In contrast, amplified J�4 and J�5products were detected when the B cells were cocultured

Figure 4. Necessity of BAFF for recombination-activating gene (RAG) induction in B cells fromhealthy controls cocultured with rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS). A,RAG-1 and RAG-2 mRNA was either amplified by reverse transcription–polymerase chainreaction from normal B cells before and after a 2-day coculture with RA FLS, cultured insupernatants from the RA FLS, or cultured in Transwell chambers apart from RA FLS. Data forcells from patients RA3 and RA5 represent independent experiments on each sample of RA FLSusing B cells from 3 different healthy controls. The 697 pre–B cell line served as a positive controlfor RAG gene expression throughout the study. B, CD40L and BAFF expression on RA FLSmembrane and BAFF expression on osteoarthritis (OA) FLS membrane were visualized (5experiments) by indirect immunofluorescence (left) and by fluorescence-activated cell sorting(right). Broken line indicates isotype control. FITC � fluorescein isothiocyanate. (Originalmagnification � 100.) C, RAG-1 and RAG-2 mRNA was amplified from healthy control B cellsbefore and after a 2-day coculture with RA FLS or OA FLS in the absence or presence of 5 �g/mlof anti-BAFF or 5 �g/ml of anti–B lymphocyte stimulator receptor 3 (anti–BR-3). Data on patientsRA6 and RA7 represent independent experiments using B cells from 4 different healthy controlsand RA FLS from 8 different patients; data for patient OA5 represent independent experimentsusing B cells from 3 different healthy controls and OA FLS from 3 different patients.

1266 ROCHAS ET AL

with RA FLS (Figure 3D), while none could be detectedwith OA FLS (Figure 3D). These products were se-quenced and were found to align with the upstream 3�end of genomic J�4 and J�5, confirming the specificity ofthe products. This observation established that not onlydo RA FLS induce the transcription of the 2 RAGgenes, but they also promote the synthesis of proteinsthat set off secondary rearrangements in B cells.

Necessity of direct cell–cell contact for the induc-tion of secondary Ig V-gene rearrangements. Require-ment of cell–cell interactions for RAG expression. RAGgenes were expressed in normal B cells upon coincuba-tion with RA FLS (Figure 4A). In contrast, incubation inthe supernatant collected from RA FLS cultured aloneexerted no effects (Figure 4A). Consistent with this needfor cell–cell contact was the observation that there wasno RAG up-regulation when normal B cells were placedin the upper chamber with RA FLS in the lowerchamber of Transwells (Figure 4A).

Dependence of RAG up-regulation on BAFF. Inprevious studies, we established that BCR ligation re-quires CD40 engagement to promote RAG gene expres-

sion in B cells (22,23). Therefore, it is possible that thecell–cell signal delivered by the RA FLS to normal Bcells is mediated by CD40L. However, fluorescencemicroscopy and FACS analysis could not detect CD40Lon FLS (Figure 4B).

Another signal involved in the induction of RAGgene expression might be BAFF. This proposition isbased on the finding that ST from RA patients expresshigh levels of BAFF (29). We first verified that BAFFwas detectable on the membrane of RA FLS (Figure4B). Further, we substantiated the need for BAFF by theinability to detect RAG when normal B cells werecocultured with RA FLS that had been preincubatedwith anti-BAFF antibodies (Figure 4C), which did notaffect viability of the cells (data not shown). Similarabrogation of RAG gene expression was obtained whennormal B cells were incubated with anti–BR-3 antibodyprior to their coculture with RA FLS (Figure 4C). NoRAG gene expression could be detected in analogouscocultures with OA FLS (Figure 4C).

To further show that BAFF is indispensable, wegenerated BAFF-deficient RA FLS by shRNA geneinterference. Survival of the RA FLS was not affected bytransfection in 5 experiments performed (data notshown). To confirm the predicted decrease in BAFFexpression, RA FLS were permeabilized, and the MFI ofBAFF expression measured by FACS. Following trans-fection with BAFF shRNA, the percentages of BAFF�cells fell from a mean � SD of 91.5 � 9.1% to 51.4 �3.4% (P � 0.05), and their MFI for BAFF expression fellfrom a mean � SD of 31.0 � 3.4 to 7.2 � 1.6 (P � 0.05)(representative example in Figure 5A). In contrast,positivity for BAFF remained stable at 92.3 � 8.2%,with an MFI of 32.5 � 2.6, after transfection with controlshRNA (Figure 5A). The implication for this finding isthat by suppressing BAFF, RA FLS lost their ability toinduce RAG in normal B cells (Figure 5B).

Requirement of a soluble factor(s) for RAG geneup-regulation. Insufficiency of BAFF for RAG gene up-regulation. In our model, BAFF alone was not sufficientto induce RAG gene expression. Although RA FLS andOA FLS expressed similar levels of BAFF (Figure 4B),OA FLS had no effects on RAG gene expression.

Thus, in an attempt to understand why OA FLSare unable to promote RAG, we compared their pheno-type with that of RA FLS. Given the overall lack ofdifferences (results not shown) between RA FLS andOA FLS with respect to expression of CD40, CD40L,and the fibroblast marker CD90 (30), we reasoned thatthe additional factor in the cooperation with transmem-brane BAFF is likely to be a soluble factor. Logically,

Figure 5. Lack of induction of recombination-activating gene (RAG)expression in B cells from healthy controls using BAFF-deficientfibroblast-like synoviocytes (FLS) from rheumatoid arthritis (RA)patients. A, BAFF expression in permeabilized RA FLS was measuredby fluorescence-activated cell sorting before (broken line) and after(solid line) transfection with BAFF short hairpin RNA (shRNA) (left)or control shRNA (right). Results are representative of 5 experiments.Dotted line indicates isotype control. MFI � mean fluorescenceintensity; FITC � fluorescein isothiocyanate. B, RAG-1 and RAG-2mRNA was amplified in normal B cells before and after a 2-daycoculture with RA FLS transfected with BAFF shRNA or controlshRNA. Data on patients RA8 and RA9 represent independentexperiments with 4 different healthy controls and RA FLS samples.The 697 pre–B cell line served as a positive control for RAG geneexpression throughout the study.

BAFF AND IL-6 FROM RA SYNOVIOCYTES INDUCE RAG IN B CELLS 1267

this factor would be secreted by RA FLS but not by OAFLS.

BAFF-induced expression of RAG in the presenceof a soluble factor. The expression of RAG was sup-pressed when RA FLS were treated with mitomycin Cbefore they were cocultured with B cells (Figure 6A).This inhibition indicates that RA FLS should be thesource of the required soluble factor. The likelihood ofthe involvement of an RA FLS–derived factor wasconfirmed by experiments in which incubation of OAFLS conditioned with supernatants from the RA FLSinduced RAG gene expression in B cells (Figure 6B).

To exclude the possibility that soluble BAFF

released from RA FLS into the media induced RAGgene expression in the latter experiments, we performed2 sets of experiments. First, we measured, but could notdetect, soluble BAFF in the conditioning supernatants(27). Second, the addition of soluble recombinant BAFFdid not enable OA FLS to up-regulate RAG geneexpression in the cocultured B cells (Figure 6B). Thesefindings confirmed that the functional BAFF contribu-tion occurs when it is in its transmembrane form.Furthermore, the facts that activated FLS produceproinflammatory cytokines and that some of them pro-mote the expression of RAG (31), prompted us to try toidentify the cytokine produced by RA FLS that assists

Figure 6. Role of interleukin-6 (IL-6) in the induction of recombination-activating gene (RAG)expression in B cells from healthy controls by rheumatoid arthritis (RA) fibroblast-like synoviocytes(FLS). A, B cells from 2 healthy controls were cocultured for 2 days with RA FLS left untreated ortreated with mitomycin C. RAG mRNA was then amplified before and after culture. Data onpatients RA8 and RA9 represent 4 independent experiments using different donors and RA FLSsamples. The 697 pre–B cell line served as a positive control for RAG gene expression throughoutthe study. B, RAG-1 and RAG-2 mRNA was amplified from B cells from healthy controls beforeand after 2 days of coculture with osteoarthritis (OA) FLS alone, with 1 �g/ml of recombinantBAFF (rBAFF), in the supernatant (sup) from cocultured FLS from patient RA1 alone or in thepresence of 40 ng/ml of anti–IL-6 receptor (anti–IL-6R) antibody, with 4 ng/ml of IL-6 alone, withIL-6 plus 40 ng/ml of anti–IL-6R antibody, with 4 ng/ml IL-4, or with granulocyte–macrophagecolony-stimulating factor (GM-CSF). Data on patient OA2 represent independent experimentswith 3 different healthy controls and OA FLS samples. C, IL-6 was assayed in supernatants from3 OA and 3 RA FLS samples by enzyme-linked immunosorbent assay with RPMI 160 medium asa negative control. Values are the mean and SD. D, RAG-1 and RAG-2 mRNA was amplified fromhealthy control B cells before and after a 2-day coculture with RA FLS alone, with 40 ng/ml ofanti–IL-6 antibody, or with 40 ng/ml of anti–IL-6R antibody. Data on patient RA3 representindependent experiments with 4 different donors and 8 RA FLS samples.

1268 ROCHAS ET AL

transmembrane BAFF in the up-regulation of RAGgenes.

Requirement of IL-6 for RAG gene up-regulation.Based on numerous observations, IL-6 has emerged asthe most credible candidate in RAG gene up-regulation.Among these observations are our previous studiesshowing that IL-6 contributes to the expression of RAGin mature B cells (23,24). Other investigators haveshown that the levels of IL-6 are elevated in the synovialfluid and serum of RA patients (32). Both in our presentstudy and in a study reported by other investigators (6),RA FLS were found to produce far more IL-6 than didOA FLS (P � 0.05) (Figure 6C). IL-6–deficient micehave been reported to be resistant to autoimmunity (33).Moreover, it was shown that blockade of IL-6 amelio-rates RA (34). In an effort to narrow down the candi-dates for the soluble factor, 2 other cytokines wereselected: IL-4, based on its definition as a cofactor forRAG expression (35), and GM-CSF, based on its inde-pendence of the proinflammatory and antiinflammatorygroups (36).

Both RAG genes were induced in B cells whencocultures with OA FLS were supplemented with IL-6(Figure 6B). Conversely, the expression of mRNA forRAG genes was blocked by anti–IL-6 antibody in normalB cells cocultured with either IL-6–supplemented OAFLS (Figure 6B) or with OA FLS conditioned withsupernatants from RA FLS (Figure 6B). Similar resultswere observed using anti–IL-6 antibody as well as anti–IL-6R antibody in normal B cells cocultured with RAFLS (Figure 6D). The results of these experimentssuggest that IL-6 is necessary as a supplementary factorfor the induction of RAG gene expression. IL-4 andGM-CSF had no effects on the expression of RAG(Figure 6B).

Overall, these results demonstrate that the en-gagement of transmembrane BAFF in FLS by BR-3 onthe surface of B cells is necessary, but not sufficient, toinduce RAG gene expression. The assistance of IL-6,which is abundantly produced by RA FLS but not by OAFLS, appears to be indispensable for this process.

DISCUSSION

In this study, we have shown that the ST micro-environment in RA, but not OA, is conducive to RAG-1and RAG-2 gene up-regulation in B cells. This concur-rent expression creates a potential for Ig V-gene revisionin activated ST B cells. Such revision may result in thegeneration of new antibody specificities, thus increasingdiversity and allowing for new anti-self specificities to

emerge. Neither RAG gene is normally expressed inblood B cells (37), yet both are expressed in B cells fromSLE patients (19,24). The current study established thattranscripts of RAG genes are absent in blood B cellsfrom patients with RA but are present in their ST B cells(17). The factors that underpin the difference in RAGgene expression between blood and ST B cells areunclear. The presence of activated B cells and germinalcenter–like structures within the pannus (38) may play arole in the up-regulation of RAG expression, akin to thereactions identified in secondary lymphoid organs (14).Our study provides a new clue as to the importance ofthe FLS for the response of B lymphocytes that infiltratethe RA ST. Through their interaction with RA FLS, Blymphocytes can be rescued from apoptosis (39,40).Thus, RA FLS are better endowed than OA FLS withthe ability to support B lymphocyte survival. As wereveal in this study, one way for self-reactive B cells toevade apoptosis within the RA ST may be by receptorrevision induced by stimulation of B cells.

Since Ig V-gene revision is known to occur onlyunder special circumstances, it was relevant to deter-mine whether RA FLS are involved in the transcriptionof RAG genes in B cells. We developed a model systemof short-term coculture of FLS from RA or OA patientswith allogeneic blood B cells from healthy controls. Thefact that B cells were induced to express RAG by RAFLS, but not by OA FLS, suggests that appropriate,potent, and persistent stimuli for V(D)J revision aregenerated by RA FLS. This process, which is likelyassociated with CD5 expression (22), may lead to theemergence of auto-reactive B cells. However, this re-mains to be demonstrated. OA FLS also express BAFFon their membrane and release minute amounts of IL-6(ref. 32 and Figure 6C in the present study). As aconsequence, B cells may expand clonally and hypermu-tate somatically in the presence of OA FLS (41).

Our findings that the 40–65% reduction in BAFFprotein expression abrogated the effect of RA FLS andthat blocking antibody to BAFF or to BR-3 preventedV(D)J gene revision highlights the importance of BAFFin this process. The exact nature and dynamics ofcell–cell contact between RA FLS and B cells warrantsfurther study. In any event, the prime physiologic acti-vator of BR-3 must be transmembrane BAFF, ratherthan its soluble form. The rationale might be thatclustering of BR-3 by transmembrane BAFF recruitsadaptors and downstream signaling partners that am-plify BAFF-induced signals. There are 2 models insupport of this proposition. The first model is suggestedby the observation that the soluble form of Fas ligand

BAFF AND IL-6 FROM RA SYNOVIOCYTES INDUCE RAG IN B CELLS 1269

has significantly reduced activity compared with itstransmembrane form in the induction of apoptosis (42).The second (43) is based on the difference betweenmechanisms of tolerance induction in B cells in micetransgenic for soluble, as compared with membrane, henegg lysozyme. Thus, whereas the former favors anergy ofspecific B cells, the latter induces apoptosis (44).

Our study further revealed that in the culturesystem we used, 2 distinct signals complement eachother to induce V(D)J gene revision. The effects onV(D)J gene revision promoted by transmembraneBAFF was not sufficient and required the concurrentpresence of IL-6. Thus, the inability for OA FLS toinduce RAG is abolished by the addition of IL-6 at theconcentration found in RA FLS supernatants (Figure6B). Also consistent with our interpretation is that IL-6was previously shown to be involved in RAG genetranscription in human lymphoid progenitor cell lines(45). Like ST B cells, bone marrow B cells need anadditional signal from stromal cells to induce their RAGgenes. In this respect, it is intriguing to note that stromalcells also express transmembrane BAFF (46).

The mechanism by which IL-6 provides the nec-essary secondary signal remains to be elucidated. IL-6may activate factors that trigger the transcription ofRAG genes (47), or it may give rise to factors that pushcis elements into counteracting the silencer that nor-mally represses RAG gene transcription (48). Morelikely, IL-6 may increase the expression level of BR-3, aswas recently demonstrated in cytokine-stimulated B cells(49), or it may help to reduce receptor numbers occu-pied by soluble BAFF to favor transmembrane BAFFengagement (50).

In conclusion, this study raises a key issue inhighlighting the ability of BAFF to induce V(D)J generearrangement in the presence of IL-6. Normally, the Bcell repertoire receives signals from T cells and cytokinesto develop into high-affinity autoantibody-secreting Bcell clones. The production of BAFF and IL-6 at sites ofinflammation in RA may therefore foster autoreactive Bcell expansion. Such uncontrolled ongoing V(D)J rear-rangement in synovial tissues could account for theexcessive autoantibody production (51) and might re-flect, at least in part, the failure of B cell toleranceobserved in RA patients (52), a hypothesis worthy offurther investigation.

ACKNOWLEDGMENTS

Thanks are due to Dr. Paul Guglielmi for the generousgift of reagents, to Professors Dominique Le Nen, Frederic

Dubrana, and Christian Lefevre for performing the surgeries,and to Christelle Le Dantec and Jean-Francois Seite for experttechnical assistance. The superb secretarial help of Cindy Seneand Simone Forest is appreciated.

AUTHOR CONTRIBUTIONS

Dr. Youinou had full access to all of the data in the study andtakes responsibility for the integrity of the data and the accuracy of thedata analysis.Study design. Hillion, Jamin, Devauchelle.Acquisition of data. Rochas, Hillion.Analysis and interpretation of data. Rochas, Hillion, Mageed, Youi-nou, Jamin, Devauchelle.Manuscript preparation. Saraux, Mageed, Youinou, Jamin, Devauchelle.

REFERENCES

1. Saraux A, Guedes C, Allain J, Devauchelle V, Valls I, Lamour A,et al. Prevalence of rheumatoid arthritis and spondyloarthropathyin Brittany, France. J Rheumatol 1999;26:2622–7.

2. Schroder AE, Greiner A, Seyfert C, Berek C. Differentiation of Bcells in the nonlymphoid tissue of the synovial membrane ofpatients with rheumatoid arthritis. Proc Natl Acad Sci U S A1996;93:221–5.

3. Youinou PY, Morrow JW, Lettin AW, Lydyard PM, Roitt IM.Specificity of plasma cells in the rheumatoid synovium. ScandJ Immunol 1984;20:307–15.

4. Reparon-Schuijt CC, van Esch WJ, van Kooten C, SchellekensGA, de Jong BA, van Venrooij WJ, et al. Secretion ofanti–citrulline-containing peptide antibody by B lymphocytes inrheumatoid arthritis. Arthritis Rheum 2001;44:41–7.

5. Chu CQ, Field M, Feldmann M, Maini RN. Localization of tumornecrosis factor � in synovial tissues and at the cartilage–pannusjunction in patients with rheumatoid arthritis. Arthritis Rheum1991;34:1125–32.

6. Guerne PA, Zuraw BL, Vaughan JH, Carson DA, Lotz M.Synovium as a source of IL-6 in vitro: contribution to local andsystemic manifestations of arthritis. J Clin Invest 1989;83:585–92.

7. Weyand CM, Goronzy JJ, Takemura S, Kurtin PJ. Cell-cellinteractions in synovitis: interactions between T cells and B cells inrheumatoid arthritis. Arthritis Res 2000;2:457–63.

8. Litinskiy MB, Nardelli B, Hilbert DM, He B, Schaffer A, Casali P,et al. DCs induce CD40-independent Ig class switching throughBLyS and APRIL. Nat Immunol 2002;3:822–9.

9. Oren DA, Li Y, Volovik Y, Morris TS, Dharia C, Das K, et al.Structural basis of BLyS receptor recognition. Nat Struct Biol2002;9:288–92.

10. Cancro MP. The BLyS family of ligands and receptors: anarchetype for niche-specific homeostatic regulation. Immunol Rev2004;202:237–49.

11. Shimaoka Y, Attrep JF, Hirano T, Ishihara K, Suzuki R, ToyosakiT, et al. Nurse-like cells from bone marrow and synovium ofpatients with rheumatoid arthritis promote survival and enhancefunction of human B cells. J Clin Invest 1998;102:606–18.

12. Meffre E, Davis E, Schiff C, Cunningham-Rundles C, Ivashkiv LB,Staudt LM, et al. Human B cells express light chains and editedreceptors. Nat Immunol 2000;1:207–13.

13. Wardemann H, Hammersen J, Nussenzweig MC. Human autoan-tibody silencing by immunoglobulin light chains. J Exp Med2004;200:191–9.

14. Giachino C, Padovan E, Lanzavecchia A. Re-expression of RAG1and RAG2 genes and evidence for secondary rearrangements inhuman germinal center B lymphocytes. Eur J Immunol 1998;28:3506–13.

15. Meffre E, Papavasiliou F, Cohen P, de Bouteiller O, Bell D,

1270 ROCHAS ET AL

Karasuyama H, et al. Antigen receptor engagement turns off theV(D)J recombination machinery in human tonsil B cells. J ExpMed 1998;188:765–72.

16. Oettinger MA, Schatz DG, Gorka C, Baltimore D. RAG1 andRAG2, adjacent genes that synergistically activate V(D)J recom-bination. Science 1990;248:1517–23.

17. Zhang Z, Wu X, Limbaugh BH, Bridges SL Jr. Expression ofrecombination-activating genes and terminal deoxynucleotidyltransferase and secondary rearrangement of immunoglobulin �light chains in rheumatoid arthritis synovial tissue. ArthritisRheum 2001;44:2275–84.

18. Itoh K, Meffre E, Albesiano E, Farber A, Dines D, Stein P, et al.Ig heavy chain variable region gene replacement as a mechanismfor receptor revision in rheumatoid arthritis synovial tissue Blymphocytes. J Exp Med 2000;192:1151–64.

19. Girschick HJ, Grammer AC, Nanki T, Vazquez E, Lipsky PE.Expression of recombination activating genes 1 and 2 in peripheralB cells of patients with systemic lupus erythematosus. ArthritisRheum 2002;46:1255–63.

20. Dorner T, Foster SJ, Farner NL, Lipsky PE. Ig � chain receptorediting in SLE. J Clin Invest 1998;102:688–94.

21. Dorner T, Farner NL, Lipsky PE. Ig � and heavy chain gene usagein early untreated SLE suggests intensive B cell stimulation.J Immunol 1999;163:1027–36.

22. Hillion S, Saraux A, Youinou P, Jamin C. Expression of RAG inperipheral B cells outside germinal centers is associated with theexpression of CD5. J Immunol 2005;174:5553–61.

23. Hillion S, Dueymes M, Youinou P, Jamin C. IL-6 contributes tothe expression of RAG in human mature B cells. J Immunol2007;179:6790–8.

24. Hillion S, Garaud S, Devauchelle V, Bordron A, Berthou C, YouinouP, et al. IL-6 is responsible for aberrant BCR-mediated regulation ofRAG expression in SLE. Immunology 2007;122:371–80.

25. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS, et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988;31:315–24.

26. Zimmermann T, Kunisch E, Pfeiffer R, Hirth A, Stahl HD, Sack U,et al. Isolation and characterization of rheumatoid arthritis synovialfibroblasts from primary culture—primary culture cells markedlydiffer from fourth-passage cells. Arthritis Res 2001;3:72–6.

27. D’Arbonneau F, Pers JO, Devauchelle V, Pennec Y, Saraux A,Youinou P. BAFF-induced changes in B cell antigenreceptor–containing lipid rafts in Sjogren’s syndrome. ArthritisRheum 2006;54:115–26.

28. Nagafuchi H, Yoshikawa H, Takeba Y, Nara K, Miura K, Kuro-kawa MS, et al. RAGs induce secondary Ig gene rearrangementand subsequent apoptosis of human circulating B lymphocytes.Clin Exp Immunol 2004;136:76–84.

29. Ohata J, Zvaifler NJ, Nishio M, Boyle DL, Kalled SL, Carson DA,et al. Fibroblast-like synoviocytes of mesenchymal origin expressfunctional BAFF in response to proinflammatory cytokines. J Im-munol 2005;174:864–70.

30. Koumas L, Smith TJ, Feldon S, Blumberg N, Phipps RP. Thy-1expression in human fibroblast subsets defines myofibroblastic orlipofibroblastic phenotypes. Am J Pathol 2003;163:1291–300.

31. Muraguchi A, Tagoh H, Kitagawa T, Nagata T, Kishi H. Stromalcells and cytokines in the induction of recombination activatinggene (RAG) expression in a human lymphoid progenitor cell.Leuk Lymphoma 1998;30:73–85.

32. Houssiau FA, Devogelaer JP, Van Damme J, Nagant de Deux-chaisnes C, Van Snick J. Interleukin-6 in synovial fluid and serumof patients with rheumatoid arthritis and other inflammatoryarthritides. Arthritis Rheum 1988;31:784–8.

33. Alonzi T, Fattori E, Lazzaro D, Costa P, Probert L, Kollias G, etal. IL-6 is required for the development of collagen-inducedarthritis. J Exp Med 1998;187:461–8.

34. Choy EH, Isenberg DA, Garrood T, Farrow S, Ioannou Y, Bird H,et al. Therapeutic benefit of blocking interleukin-6 activity with ananti–interleukin-6 receptor monoclonal antibody in rheumatoidarthritis: a randomized, double-blind, placebo-controlled, dose-escalation trial. Arthritis Rheum 2002;46:3143–50.

35. Kajiwara K, Shinazawa M, Morishima H, Yanagihara Y. Differ-ential effect of IL-4 and IL-13 on the expression of recombination-activating genes in mature B cells from human peripheral blood.Cell Immunol 2004;227:121–8.

36. Till KJ, Burthem J, Lopez A, Cawley JC. Granulocyte-macrophagecolony-stimulating factor receptor: stage-specific expression andfunction on late B cells. Blood 1996;88:479–86.

37. Girschick HJ, Grammer AC, Nanki T, Mayo M, Lipsky PE. RAG1and RAG2 expression by B cell subsets from human tonsil andperipheral blood. J Immunol 2001;166:377–86.

38. Goronzy JJ, Weyand CM. Rheumatoid arthritis. Immunol Rev2005;204:55–73.

39. Dechanet J, Merville P, Durand I, Banchereau J, Miossec P. Theability of synoviocytes to support terminal differentiation of acti-vated B cells may explain plasma cell accumulation in rheumatoidsynovium. J Clin Invest 1995;95:456–63.

40. Lindhout E, van Eijk M, van Pel M, Lindeman J, Dinant HJ, deGroot C. Fibroblast-like synoviocytes from RA patients haveintrinsic properties of follicular dendritic cells. J Immunol 1999;162:5949–56.

41. Da RR, Qin Y, Baeten D, Zhang Y. B cell clonal expansion andsomatic hypermutation of Ig variable heavy chain genes in thesynovial membrane of patients with osteoarthritis. J Immunol2007;178:557–65.

42. Holler N, Tardivel A, Kovacsovics-Bankowski M, Hertig S, GaideO, Martinon F, et al. Two adjacent trimeric Fas ligands arerequired for Fas signaling and formation of a death-inducingsignaling complex. Mol Cell Biol 2003;23:1428–40.

43. Goodnow CC, Crosbie J, Jorgensen H, Brink RA, Basten A.Induction of self-tolerance in mature peripheral B lymphocytes.Nature 1989;342:385–91.

44. Hartley SB, Crosbie J, Brink R, Kantor AB, Basten A, GoodnowCC. Elimination from peripheral lymphoid tissues of self-reactiveB lymphocytes recognizing membrane-bound antigens. Nature1991;353:765–9.

45. Tagoh H, Kishi H, Okumura A, Kitagawa T, Nagata T, Mori K, etal. Induction of RAG expression in a human lymphoid progenitorcell line: requirement of two separate signals from stromal cellsand cytokines. Blood 1996;88:4463–73.

46. Gorelik L, Gilbride K, Dobles M, Kalled SL, Zandman D, Scott ML.Normal B cell homeostasis requires B cell activation factor produc-tion by radiation-resistant cells. J Exp Med 2003;198:937–45.

47. Monroe RJ, Chen F, Ferrini R, Davidson L, Alt FW. RAG2 isregulated differentially in B and T cells by elements 5� of thepromoter. Proc Natl Acad Sci U S A 1999;96:12713–8.

48. Yannoutsos N, Barreto V, Misulovin Z, Gazumyan A, Yu W,Rajewsky N, et al. A cis element in the RAG locus regulates geneexpression by counteracting a distant silencer. Nat Immunol2004;5:443–50.

49. Darce JR, Arendt BK, Wu X, Jelinek DF. Regulated expression ofBAFF-binding receptors during human B cell differentiation.J Immunol 2007;179:7276–86.

50. Carter RH, Zhao H, Liu X, Pelletier M, Chatham W, Kimberly R,et al. Expression and occupancy of BAFF-R on B cells in systemiclupus erythematosus. Arthritis Rheum 2005;52:3943–54.

51. Dorner T, Lipsky PE. Immunoglobulin variable-region gene usagein systemic autoimmune diseases [review]. Arthritis Rheum 2001;44:2715–27.

52. Samuels J, Ng YS, Coupillaud C, Paget D, Meffre E. Human B celltolerance and its failure in rheumatoid arthritis. Ann N Y Acad Sci2005;1062:116–26.

BAFF AND IL-6 FROM RA SYNOVIOCYTES INDUCE RAG IN B CELLS 1271